- © Fotolia
Objectives: Demonstrate - through life-size platforms on complex systems - the unprecedented width spectrum of the site’s digital skills and its multiple academic and socio-economic interconnections.
Under the direction of Marius Tucsnak
The SysNum cluster is a follow-up of the CPU cluster, which aimed to bring Bordeaux academic players in the digital sciences closer to each other, to increase their know-how and to make them diffuse both within and outside the academic world.
With the SysNum cluster, a new stage has been achieved in terms of structuring, since the community is joining forces around large-scale distributed digital systems to give shape to 3 application platforms around major societal issues: ecology, mobile systems, interconnected objects and data analysis.
The key words are:
With 4 research axes:
- Interconnected object systems
- Reliability and safety
- Modeling and numerical systems
- Massive and heterogeneous data
With 3 demonstrators:
- Smart ecological systems
- Robotics and drones
- Smart campus
Objectives: unite key Bordeaux public stakeholders in the digital sciences to develop these to a level allowing their use as a certification tool; stimulate the innovation, competitiveness, visibility and attractiveness of the CPU community in terms of research, education and development, especially internationally.
Under the leadership of Thierry Colin
The project was completed by the end of 2016
The idea of using scientific computation as a certification tool, or at the very least as a qualification tool, is progressively gaining support, not only because the power of computers is increasing continuously, but also on the strength of the methods and models developed over the past decades.
In order to deal with the problems in the fields of certification or qualification, a combination of several scientific disciplines is required: mathematical and mechanical modeling, digital analysis and computer science.
The use of thousands of processors is useless without reliable digital methods or accurate models.
In Bordeaux, a critical mass of researchers are currently working on these issues, both in theory and in practice, within the framework of an established interdisciplinary process.
The researchers in the CPU project are focusing their energy on three main areas of research leading to major advances in several fields of application which require high reliability:
- Land transport
|3 areas: Scientific computing; Signal-Image; Reliability, safety, decision-making||approximately 220 members (lecturers, researchers, engineers, etc.)||43 research projects with CPU funds|
|8 research projects with IdEx funds involving the CPU (3 from the InterLabEx program, 5 from the CPU transfer program)||4 partnerships with major international universities||backing of 5 training courses at University of Bordeaux and Bordeaux INP|
|3 start-up projects maturing|
In the field of IT, nothing should crash or freeze. If this were the case, it is definitely due to human error. To overcome such issues, we have set up a software platform for visualization and simulation of distributed algorithms, Visidia; a platform that has been approved within the framework of CPU. It relies on visualization to understand the basics of distributed computing. When you write an algorithm, it is sequential in your head, you write for one process at a time. But today, once you are behind a computer, you do not address a single user, but a group connected by the network, up to several million at the same time. You have to think differently.
Mohamed Mosbah - Deputy director of LaBRI - Bordeaux INP
- Automatic segmentation of 3D MRI obtained with volBrain - © Pierrick CoupevolBrain: an online MRI brain volumetry system
Pierrick Coupé (LaBRI - University of Bordeaux, CNRS, France) - a member of the CPU cluster - and José V. Manjón (IBIME - UPV, Spain) have developed volBrain : a new, free online system able to automatically analyze MRI data from the brain. The system enables scientists worldwide to obtain key cerebral information in order to advance in research on neural pathologies. It provides information on the tissue volume in the intracranial cavity, as well as the cerebral hemispheres, the cerebellum and the brain stem.